Change the situation, run the same decision on both sides, and inspect the shape of each answer. Four browser-only simulations; no keys or login needed.
Browser simulation · no API calls
01 / Shape the state
Same question to both
Which queue should receive this ticket?
02 / Inspect the contract
Billing is the best route. The customer appears time-sensitive and mentions a charge. I would send this to the billing team and include a short summary for the agent.
Requested JSON
{ "route": "billing", "reason": "best semantic match" }Flexible response parse what you need
Decision
Billing
Confidence
90%
Fixed primitive consume directly
What to notice
The downstream system needs one queue name. Jev's bounded Choice is the direct interface; the LLM's explanation is useful only if a human also needs context.
Need prose? Generate with an LLM.
Need a bounded decision? Test a Jev primitive.
LLMs generate
A general-purpose LLM can explain, rewrite, and produce flexible text. You can ask it for JSON, but the task is still framed as generation.
Jev chooses
Jev is designed around Choice, Score, and Noul: bounded outputs for repeated decisions where the allowed shape is known in advance.
Fit decides
Use a bounded primitive when the next system step needs a label or probability. Use an LLM when a person needs prose, nuance, or synthesis.
50 runnable Jev use cases
The open-source gallery this practice lab is based on.
How Jev works
Choice, Score, Noul, and the System One model idea.
Where Jev actually fails
Why valid structure is not the same as a correct decision.
Jev for prompt-injection triage
A deeper look at the security-gate use case.
No. It is a deterministic teaching simulation that runs entirely in your browser. The labels, probabilities, explanations, and output sizes are illustrative examples, not API measurements.
No. It compares output shapes, not accuracy, speed, or price. Both approaches can make wrong decisions, and production evaluation requires your own representative data.
Choice selects from fixed options, Score returns a bounded probability-weighted value, and Noul returns a yes-or-no probability. They are the three Jev primitives represented in this lab.
Use one when the output needs open-ended language, synthesis, transformation, or a response for a person. A bounded decision can also route into an LLM that handles the prose afterward.
Tokenizer Playground
Type anything and watch it split into tokens live — the units LLMs actually read and get billed for.
Context Window Visualizer
See how much of a model's context window your text fills up, and what happens when you run out of room.
FROG in a Bowl Prompt Builder
Fill in Format, Role, Objective, Goal, and Context — get a copy-ready structured prompt in seconds.
Machine Learning Types
Three tiny games show what supervised, unsupervised, and reinforcement learning actually mean.
Neural Network Playground
Drag two sliders and watch a real, tiny neural network turn them into a decision, live.
Generative AI Playground
Watch AI write one word at a time by predicting what's most likely to come next.
RAG Playground
Ask a question, watch notes get retrieved, then see a grounded answer versus a hallucination.
Embedding Map
Click two words and see why similar meanings sit near each other — the idea behind vector search.
Attention Visualizer
Click a word and see which others a toy transformer looks at — including the classic “it” puzzle.
Prompt Injection Lab
Watch a pasted email try to override a support agent, then flip a switch that treats it as data.